Bicyclists’ avoidance behavior around static and dynamic obstacles: A trajectory-based analysis using drone-recorded data

Introduction: Bicycle safety is a critical concern in urban transportation, particularly when bicyclists encounter obstacles. This study investigates bicyclists’ avoidance behavior in five common obstructed scenarios—including three static obstacles and two dynamic obstacles—using 1177 validated cases derived from drone-collected trajectory data. Method: We analyze three motion characteristics (velocity, acceleration, deceleration) and introduce two novel safety metrics: acceptable risk, based on risk field theory, and acceptable distance, calculated via convex polygon geometry. Results: The results indicate that the motion state of an obstacle, rather than its type or quantity, serves as the primary determinant of bicyclists’ avoidance behavior. When encountering dynamic obstacles, bicyclists significantly reduce speed by 1.04 m/s (3.43 ft/s) (p < 0.01), increase acceleration by 0.35 m/s 2 (1.16 ft/s 2 ) (p < 0.01), increase deceleration by 0.43 m/s 2 (1.42 ft/s 2 ) (p < 0.01), and maintain a greater acceptable distance by 0.23 m (0.76 ft) (p < 0.01), compared to static obstacles. Furthermore, bicyclists’ acceptable risk (mean = 0.602, SD = 0.078) remains consistent across scenarios, indicating that bicyclists maintain a stable risk acceptance margin across different obstacles. However, bicyclists exhibit a 14% higher risk tolerance compared to car drivers under identical conditions (p < 0.01). Additional analysis reveals that bicyclists with higher acceptable risk (labeled as aggressive) exhibit higher speed by 0.60 m/s (1.98 ft/s), acceleration by 0.12 m/s 2 (0.38 ft/s 2 ) and deceleration by 0.09 m/s 2 (0.30 ft/s 2 ) than conservative bicyclists with lower acceptable risk (labeled as conservative). Conclusions and Practical Applications: These findings establish a foundational understanding of bicyclists’ avoidance behavior, thereby offering valuable insights for the development of traffic microsimulation models and the design of bicycling infrastructure, with positive implications for overall traffic safety.